How to write a proposed endeavor

If you get one thing right in an EB-2 NIW petition, make it this. The proposed endeavor is the anchor the entire case hangs from: prong 1 evaluates it, prong 2 asks whether you can advance it, and prong 3 asks whether the waiver is justified for it. A vague endeavor makes all three unanswerable.

The test

If you cannot describe your endeavor in two or three sentences with enough specificity that a stranger could evaluate it, it is not defined well enough to build a petition around.

Not "explain it to an expert." A stranger. The officer reading your petition adjudicates across every field and has limited time per case.

Field level versus endeavor level

This is the distinction that decides cases.

❌ "I plan to conduct research in artificial intelligence."

Any credentialed person in the field could write that sentence. It gives an officer nothing to evaluate, and no basis to conclude the waiver is warranted for you specifically rather than anyone else with the same degree.

✅ "I am developing machine-learning models to improve early detection of antibiotic-resistant infections in hospital settings, where current protocols fail and mortality is highest."

Now there is something to assess: a named problem, a named population, a stated failure in the status quo, and an implied measure of success.

More pairs, all reflecting real distinctions from published outcomes:

Field level ❌	Endeavor level ✅
"Research in materials science"	"Developing reusable, non-toxic organic substitutes for the wasteful catalysts currently standard in industrial chemical processing"
"Cybersecurity work"	"Building zero-day threat prevention using deep learning models, deployed to protect U.S. critical infrastructure and government systems"
"Teaching at the university level"	"Teaching English language and literature at U.S. universities, addressing a documented national shortage of qualified language faculty"
"Software engineering"	"Providing cloud, data-management, and scalable software solutions to U.S. small and medium enterprises, plus a talent incubator training developers to address the regional shortage of skilled IT professionals"
"Improving business efficiency"	"Improving cybersecurity standards across regional healthcare systems, protecting patient data and strengthening public trust"

Note what the right-hand column has in common. Each names a problem, a mechanism, and a beneficiary who is not the petitioner.

The four questions your endeavor must answer

From the denial patterns, in order:

1. What problem does it address? Concretely. Not "inefficiency", the specific failure.
2. Why does that problem matter to the United States? Connect it to a present national objective, not a general good.
3. Who benefits beyond you, your employer, and your clients? If the honest answer is "my clients," the endeavor needs reframing, not better wording.
4. How does the benefit extend past one employer or a small client base? Adoption, replication, standards, documented shortage, multi-state uptake.

The private-benefit trap

The most common fatal framing, and the hardest to see in your own work.

An immigration attorney was denied because her endeavor benefited her clients. A real estate developer was denied because his plan was not credible on the record. A professor was denied because the benefit stopped at her institution. None of the three lacked ability.

Ask yourself directly: who benefits who is not paying me? If you cannot name them and evidence them, prong 1 is not established, however strong the rest of your file.

Two ways out, both legitimate:

• Widen the mechanism. The software engineer in our corpus who was approved after an RFE did not change what he did, he evidenced that his work served SMEs across an industry rather than individual clients, with a Chamber of Commerce letter showing regional demand and an independent economist's analysis.
• Name the systemic effect. A tool adopted as a standard, a method others replicate, a shortage you measurably reduce.

Indirect impact needs the chain written out

Some endeavors benefit the country through several steps. That is fine, but denied.md is explicit that the officer should not have to infer the connection.

A workforce training programme improves regional employment stability → which reduces dependence on public assistance → which strengthens the regional tax base.

Write each link. Provide both a technical explanation and a plain-language one; approved petitions in our corpus did exactly that, and one case study notes the letters included "technical and layperson versions to appeal to USCIS officers."

The execution plan is part of the endeavor

A strong idea with no plan reads as an intention, and a vague execution plan is one of the most common RFE triggers. State:

• Where you begin, the concrete first step, not the vision
• The steps after it
• The resources you will use: capital, partners, institutions, infrastructure
• How it scales or expands
• The outcomes you expect, and how you would measure them
• Any licensing pathway, if your endeavor requires U.S. licensure, silence here invites questions from engineers, physicians, and architects especially

Where the endeavor includes components outside your expertise, say how they get covered. The source material gives the example directly: an engineer proposing advanced AI tools without AI experience should explain the division of responsibilities.

Can you change your endeavor after filing?

Carefully. One petitioner in our corpus received an RFE saying the endeavor was not specific, and responded by narrowing from field level to a named project, framing it explicitly as a clarification of the original description rather than a new endeavor. That worked, and they reported worrying it would be treated as a material change.

The lesson is to get it right before filing. Narrowing on an RFE is possible but it is a risk you do not need to take.

A working method

1. Write the endeavor at field level first, just to get it out.
2. Ask "which specific project?" until the answer names one thing.
3. Write the four answers above, in order, in plain prose.
4. Draft prong 3 next, why the waiver is warranted for you. If you cannot without leaning on how impressive you are, go back to step 2.
5. Only then assemble evidence. Evidence chosen before the endeavor is settled is evidence that will not match it.

Next: the three prongs in detail · what evidence actually counts · assess your record